b-bolivia, an allele of the maize b1 gene with variable expression, contains … · b-bolivia, an...

B-Bolivia, an Allele of the Maize b1 Gene with VariableExpression, Contains a High Copy Retrotransposon-Related Sequence Immediately Upstream1

David A. Selinger and Vicki L. Chandler*

Department of Plant Sciences, University of Arizona, Tucson, Arizona 85721

The maize (Zea mays) b1 gene encodes a transcription factor that regulates the anthocyanin pigment pathway. Of the b1alleles with distinct tissue-specific expression, B-Peru and B-Bolivia are the only alleles that confer seed pigmentation.B-Bolivia produces variable and weaker seed expression but darker, more regular plant expression relative to B-Peru. Ourexperiments demonstrated that B-Bolivia is not expressed in the seed when transmitted through the male. When transmittedthrough the female the proportion of kernels pigmented and the intensity of pigment varied. Molecular characterization ofB-Bolivia demonstrated that it shares the first 530 bp of the upstream region with B-Peru, a region sufficient for seedexpression. Immediately upstream of 530 bp, B-Bolivia is completely divergent from B-Peru. These sequences share sequencesimilarity to retrotransposons. Transient expression assays of various promoter constructs identified a 33-bp region inB-Bolivia that can account for the reduced aleurone pigment amounts (40%) observed with B-Bolivia relative to B-Peru.Transgenic plants carrying the B-Bolivia promoter proximal region produced pigmented seeds. Similar to native B-Bolivia,some transgene loci are variably expressed in seeds. In contrast to native B-Bolivia, the transgene loci are expressed in seedswhen transmitted through both the male and female. Some transgenic lines produced pigment in vegetative tissues, but thetissue-specificity was different from B-Bolivia, suggesting the introduced sequences do not contain the B-Bolivia plant-specificregulatory sequences. We hypothesize that the chromatin context of the B-Bolivia allele controls its epigenetic seedexpression properties, which could be influenced by the adjacent highly repeated retrotransposon sequence.

The alleles of the b1 locus of maize (Zea mays)display a high degree of phenotypic diversity interms of tissue- and developmental stage-specific ex-pression (Styles et al., 1973; Coe, 1979; Selinger andChandler, 1999). Studies on several alleles haveserved as a useful system to investigate how majorchanges in tissue-specific gene expression occurred(Radicella et al., 1992; Selinger et al., 1998; Selingerand Chandler, 1999). The b1 locus encodes a tran-scription factor that regulates anthocyanin pigmentexpression, which provides an excellent visualmarker for gene expression. The B-I and B-Peru al-leles represent the extremes of the phenotypic diver-sity of b1 alleles. B-I is highly expressed in most of thevegetative tissues of the plant but is not expressed inthe embryo or aleurone tissues of the seed. In con-trast, B-Peru is weakly and variably expressed invegetative tissues of the plant, but is highly ex-pressed in part of the embryo and in the aleuronelayer of the seed (for a detailed description of thesetwo alleles, see Radicella et al., 1992). The B-Bolivia

allele has an intermediate phenotype between thesetwo alleles. B-Bolivia, like B-Peru, pigments the aleu-rone layer of the seed and these are the only knownb1 alleles that confer aleurone-specific pigmentation.However, the consistency of pigmentation in thealeurone layer of the seed is quite different in the twoalleles (Styles et al., 1973). The plant pigmentationdirected by B-Bolivia can be as dark as that in B-I, butB-Bolivia pigments a subset of the plant vegetativetissues relative to B-I.

The ability of B-Bolivia to pigment both seed andplant tissues is reminiscent of alleles of r1. The r1gene encodes a homologous and functionally dupli-cate protein to that of b1 (Ludwig et al., 1989; Goff etal., 1990; Ludwig et al., 1990) and like b1, the r1 genehas many phenotypically diverse alleles, many ofwhich color the aleurone layer of the seed (Styles etal., 1973). Several of the r1 alleles that color both seedand plant tissues have separate coding regions thatare expressed in the seed or in the plant tissues(Stadler and Neuffer, 1953; Robbins et al., 1991;Walker et al., 1995).

Previous investigations of B-Peru and B-I havedemonstrated that each is a simple allele consistingof a single coding sequence. Characterization of thesequences responsible for the aleurone expression ofthe B-Peru allele and investigation of the phyloge-netic relationships between several b1 alleles haverevealed that distinct phenotypes correlate withrearrangements or insertions in the upstream region

1 This work was supported by a postdoctoral fellowship fromthe Jane Coffin-Childs Memorial Fund for Medical Research (toD.A.S.), by a grant from the U.S. Department of Agriculture Na-tional Research Initiative (grant no. 96 –35301–3179 to V.L.C.), andby the Department of Army Research for the purchase of theMolecular Dynamics Storm 860 system used in this work (grantno. DAAG559710102).

* Corresponding author; e-mail [email protected]; fax520 – 621–7186.

Plant Physiology, March 2001, Vol. 125, pp. 1363–1379, www.plantphysiol.org © 2001 American Society of Plant Physiologists 1363 www.plantphysiol.orgon February 26, 2020 - Published by Downloaded from Copyright © 2001 American Society of Plant Biologists. All rights reserved.

http://www.plantphysiol.org

of several alleles (Radicella et al., 1992; Selinger et al.,1998; Selinger and Chandler, 1999).

Using genetic and molecular techniques, we havecharacterized the expression and structure of theB-Bolivia allele. Our results indicate that B-Boliviacontains a single coding region and that the B-Peruand B-Bolivia alleles share most of the sequencesrequired for aleurone expression. Immediately up-stream of the aleurone-specific sequences in B-Boliviais a highly repeated retrotransposon-related se-quence. Transient transformation assays and trans-genic plants were used to characterize sequences re-quired for seed expression. Our results suggest thatthe retro-element-related sequences immediately up-stream of the aleurone-specific promoter contributesto some but not all of the epigenetic differences inseed expression between B-Peru and B-Bolivia.

RESULTS

The B-Bolivia Allele Shows Uniform Plant ExpressionBut Variable Seed Expression

The B-Bolivia allele, similar to B-I, conditions stronganthocyanin pigmentation in several vegetative planttissues (Coe, 1979), including culm, leaf sheath, andhusk tissues (Fig. 1A). B-Bolivia also directs anthocy-anin expression in the aleurone layer of the seeds. Itis the only b1 allele besides B-Peru that confers aleu-rone pigmentation. Whereas the B-Peru allele alwaysconferred uniform and intense pigmentation of thealeurone layer (Fig. 1B), aleurone pigmentation byB-Bolivia was weaker and variable (Fig. 1C). As notedby previous workers (Styles et al., 1973), we observedthree major differences relative to B-Peru. First, notevery seed that inherits B-Bolivia was pigmented.Second, the amount of pigment in different kernelsvaried, even though each kernel was homozygous forB-Bolivia. Third, the amount of pigment even in the

darkest B-Bolivia seeds was less than that conferredby B-Peru. We have further observed that the propor-tion of B-Bolivia seeds expressing pigment variedbetween different genetic stocks. The ear in Figure 1Cis illustrative of the low number of purple kernelsobserved when B-Bolivia is in the K55 genetic back-ground. The ear in Figure 1D illustrates that a largernumber of purple kernels are observed whenB-Bolivia is in a different background, in this case onederived from the George Sprague (GS) B-Bolivia line.The vegetative plant expression is equivalent in boththe K55 and GS genetic backgrounds (data notshown).

B-Bolivia Is Not Expressed in the Aleurone ifTransmitted through the Male Parent

During our characterization of B-Bolivia seed ex-pression we discovered that the presence of coloredkernels in progeny from outcrosses between B-Boliviastocks with stocks containing recessive b1 alleles de-pended on which stock was the male parent. Earsfrom B-Bolivia plants crossed by pollen from plantswith recessive b1 alleles displayed frequencies of pig-mented seeds and intensity of pigmentation that wasindistinguishable from self-pollinated ears (Fig. 1D).In contrast, ears from plants with recessive b1 allelescrossed by pollen from homozygous or heterozygousB-Bolivia plants displayed no colored kernels (Fig.1E). B-Peru, the other b1 allele with aleurone pigmen-tation does not show female-specific expression.

To further explore B-Bolivia transmission, recipro-cal crosses were performed between stocks with re-cessive b1 alleles and the K55 and GS B-Bolivia stocks.The data presented in Table I (experiment 1–4)showed that seeds were pigmented on B-Bolivia earsthat were pollinated by the b1 tester stock. However,the ears from b1 tester plants pollinated by B-Bolivia

Figure 1. Phenotypes of B-Bolivia. A, Vegetative plant pigmentation in a B-Bolivia plant: a, auricle; c, culm; and m, leafmid-vein. B, A self-pollinated ear from a plant homozygous for B-Peru. C, A self-pollinated ear from a homozygous B-Boliviaplant that shows the incomplete penetrance of seed expression typical of B-Bolivia. This ear is in the K55 background. D,This ear resulted from pollination of a heterozygous B-Bolivia/b plant by a b1 tester line that has recessive nonfunctionalalleles of b1 and r1, and functional alleles of all other genes required for anthocyanin production. Self-pollinated ears fromsibling plants were indistinguishable. E, An ear from a b1 tester plant pollinated by a heterozygous B-Bolivia/b plant. Thisear is representative of ears from the reciprocal cross that produced the ear in D.

Selinger and Chandler

1364 Plant Physiol. Vol. 125, 2001 www.plantphysiol.orgon February 26, 2020 - Published by Downloaded from Copyright © 2001 American Society of Plant Biologists. All rights reserved.


stocks produced no pigmented kernels. Monitoringplant pigmentation demonstrated that B-Bolivia istransmitted through both the male and female ga-metes. Colorless seeds from the reciprocal crossesproduced as darkly pigmented plants as coloredseeds with no consistent differences in plant pigmen-tation (data not shown).

The above results indicated that B-Bolivia was sub-ject to parent of origin-specific expression in the aleu-rone, either because of genomic imprinting or be-cause of an effect of gene dosage on B-Boliviaexpression. Because the aleurone layer is derivedfrom the triploid endosperm there is only one copypresent when B-Bolivia is transmitted through themale. In contrast, kernels on the reciprocally crossedear, in which B-Bolivia is the female parent and re-cessive b1 the male, have two copies of B-Bolivia, andkernels on a self-pollinated ear from a B-Bolivia ho-mozygote have three copies of B-Bolivia.

The classic experiment to determine whether dos-age effects or gamete transmission are responsible forexpression differences in maize seed is to increase thedosage in the male or reduce the dosage in the femaleusing genetic tools (Kermicle, 1970). Despite consid-erable effort, we were unable to use similar methodsto alter the dosage of B-Bolivia through either themale or female. To examine whether we could see adosage effect between seeds carrying two and threedoses of B-Bolivia, we compared the proportion ofpigmented seeds on ears that were self-pollinatedwith ears crossed by plants with recessive b1 allelesin two different stocks, GS and 414. The results pre-sented in Table I (experiment 4–7) showed that inboth stocks there were slight differences between theproportion of colored kernels on self-pollinated andoutcrossed ears, but this difference was not statisti-

cally significant. Further evidence that the differ-ences were not biologically significant came from theobservation that the differences in GS and 414 werein opposite directions. The GS outcross ears hadfewer colored kernels, but the 414 outcrossed earshad more colored kernels than the self-pollinatedears. The observation that kernels carrying three cop-ies of B-Bolivia were no more likely to be pigmentedthan those carrying two copies suggests that dosagedifferences are not responsible for the pigmentdifferences.

Pigmented B-Bolivia Seeds Are Not HeritablyDifferent from Colorless Seeds with Respect to theProportion of Pigmented Seeds Produced in theNext Generation

Given the variability of seed expression, we wereinterested in determining the heritability of aleuroneexpression in B-Bolivia. To test for a correlation be-tween pigmentation of the seed of a parent plant andthe proportion of pigmented seeds in its progeny, weplanted colored and colorless seeds from the samehomozygous B-Bolivia ears, self-pollinated the result-ing plants, and determined the percentage of coloredseeds in the progeny. The experiment, summarizedin Table II, was performed with two differentB-Bolivia stocks, a stock that produced low numbersof purple seeds, 1470 (the parental ear had 9% col-ored kernels), and a stock that produced increasednumbers of purple seeds, GS (the parental ears aver-aged 64% colored kernels). Plants grown from pig-mented 1470 seeds produced ears that averaged28.7% colored kernels, whereas pigmented GS seedsproduced plants with ears averaging 75.5% coloredkernels. The plants grown from colorless 1470 seeds

Table I. Parent of origin differences on B-Bolivia expression

Experiment Stock Crossa Meanb N c Highd Lowd

% %

1 K55 b13 0 36 0 02 K55 3b1 8.0 11 13.5 2.43 GS b13 0 10 0 04 GS 3b1 64.1e 11 83.6 41.15 GS Self 71.1e 9 93.3 55.96 414 3b1 12.1e 19 31.3 1.47 414 Self 9.5e 36 27.2 1.3

a The direction and identity of the other parent in the cross that generated the ears that were scoredis indicated as follows: 3b1, the B-Bolivia stock was crossed by b1 tester pollen; and b13, thereciprocal cross in which B-Bolivia pollen was placed on b1 tester ears. b The average of theproportion of colored kernels on each ear is given. Kernels were counted as pigmented if they could beclearly distinguished from colorless sibling kernels; a wide range from dark to very lightly pigmentedkernels were included. c The no. of ears that were counted and used to determine the mean. d Theproportion of colored kernels on the ears with the highest and lowest percentages in the sample. e Wecalculated the probability that the difference between the self cross and the outcross by b1 pollen(B-Bolivia 3 b1) is due to chance variation for GS and 414 samples using the ANOVA method. This testcalculates the probability that the difference in the mean of two groups of numbers is due to randomchance. The test results revealed that neither of these differences is statistically significant (GS, P 5 0.24;414, P 5 0.21).

B-Bolivia Structure and Expression

Plant Physiol. Vol. 125, 2001 1365 www.plantphysiol.orgon February 26, 2020 - Published by Downloaded from Copyright © 2001 American Society of Plant Biologists. All rights reserved.


averaged 20.9% colored kernels and those of GS av-eraged 62.1% colored kernels. In both cases, theplants grown from colored seeds produced a some-what higher proportion of colored kernels, howeverthis difference between the average values was notstatistically significant. These results indicate that theon or off pigment expression state of a particular seedis not heritable because both types produce similarnumbers of colored kernels in progeny.

The loss of pigment that occurs upon male trans-mission is also not heritable. When colorless kernelsfrom such ears as shown in Figure 1E are planted andcrossed by pollen from plants carrying a recessive b1allele, a similar number of colored kernels are ob-served when compared with kernels that derive fromonly female transmission. For example, in one exper-iment, an individual B-Bolivia/b1 plant was both self-pollinated and outcrossed as male to a b1 tester plant.Colorless seeds from both ears were planted, plantscarrying the B-Bolivia allele were determined by plantcolor, and such ears were crossed by pollen from a b1tester line. The 10 plants, which resulted from the col-orless B-Bolivia/b1 seeds that resulted from the orig-inal cross as male to the b1 tester plant, produced earsin which 46% of the kernels carrying B-Bolivia werepigmented with a range between 17% to 74%. Thiswas equivalent to the six B-Bolivia/b1 heterozygousplants derived from colorless kernels on the self-pollinated ear, as the proportion of B-Bolivia kernelsthat expressed color on these ears averaged 47% witha range from 19% to 77%.

B-Bolivia Allele Has Part of the B-Peru Aleurone-Specific Promoter Region

To investigate whether B-Bolivia is a simple or com-plex allele and to determine if its unique expressionpatterns result from unique promoter sequences, amolecular study of B-Bolivia was initiated. Initially arestriction map was generated for B-Bolivia usingDNA gel blots probed with DNA fragments derivedfrom the B-I and B-Peru alleles. Using the 550b probethat lies near the 59 end of the transcribed region in

B-I and B-Peru (Patterson et al., 1995), we found thatthere is a single b1 coding region in B-Bolivia and thatit had the same map as the B-Peru and B-I codingregions (Fig. 2A). Because B-Peru and B-I differ dra-matically in the upstream region, we extensivelymapped the region of B-Bolivia upstream of the tran-scribed sequences. We used the 550b probe and com-bined double digests with BamHI and other enzymes.BamHI cuts at the 39 end of the 550b sequence pro-viding an anchor site to facilitate mapping (Fig. 2A).We found that the upstream region of B-Bolivia con-tained many distinct RFLP when compared with thesame regions of B-Peru and B-I (Fig. 2A; data notshown).

Based on the mapping, we identified a 2.8-kbBamHI fragment from B-Bolivia for cloning, whichcontained approximately 2.1 kb of upstream se-quence (“Materials and Methods”). Two l-cloneswere isolated from a size selected BamHI digestedgenomic DNA library, converted to a plasmid andthe inserts were restriction mapped. Both inserts hadidentical restriction maps that matched the restrictionfragment sizes determined from DNA-blot analysisof genomic DNA. One of the two clones was com-pletely sequenced and this sequence was comparedwith that of the B-Peru and B-I alleles. These se-quence comparisons revealed that part of the up-stream region of the B-Bolivia clone was almost iden-tical to part of the aleurone-specific promoter ofB-Peru, differing by a single 4-bp insertion in theB-Bolivia sequence relative to B-Peru. This homologyextended to 530 bp upstream of the start of transcrip-tion (for diagram, see Fig. 2B). Beyond this point thesequences completely diverged.

Divergent Sequence in B-Bolivia Has Homology toRetrotransposons and Is Present in the MaizeGenome in Very High Copy Number

To determine the nature of the divergent sequencein B-Bolivia, sequence homology searches of GenBankwere conducted and DNA-blot analyses were per-formed using the divergent sequence as a probe.

Table II. Effect of seed color on expression in the next generationColored and colorless seeds from homozygous B-Bolivia were planted, the resulting plants selfed, and ears scored for the proportion of purple

colored kernels.

Stock Seed Colora Meanb N c Highd Lowd P e

% %

1470 Purple 28.7 3 52.2 13.2 –1470 Colorless 20.9 10 35.2 9.2 0.34GS Purple 75.5 6 93.3 55.9 –GS Colorless 62.1 3 64.4 58.2 0.19

a The color of the seeds that were planted is indicated. Purple seeds were clearly distinguishable from colorless seeds, but were not necessarilythe darkest seeds on the ear. b The average of the proportion of colored kernels on each ear is given. c The no. of ears that were countedand used to determine the mean. d The proportion of colored kernels on the ears with the highest and lowest percentages in thesample. e The probability that the difference between the ears derived from plants grown from purple and colorless seeds is due to chancevariation. The probability was determined using the ANOVA method. In both cases, the differences between the plants grown from colored andcolorless kernels were not statistically significant.




Using BLASTN (Altschul et al., 1990) to search forDNA sequences homologous to the upstream regionof B-Bolivia no significant homologies were foundoutside of the region that is nearly identical to B-Peru.Using BLASTX and FASTX searches (Altschul et al.,1990; Pearson et al., 1997), in which the nucleotidesequence of B-Bolivia is translated into all six possiblepolypeptide sequences and compared with the pro-tein sequence database, we found a sequence withinthe B-Bolivia upstream region that, when translated,had highly significant sequence identity to proteinsequences in GenBank. This search identified a read-ing frame with 35% identity over 308 amino acids (Bitscore of 129, E(513612) was 3e-27) to a gag polypro-tein from Sorghum (gb: AAD19359.1), a protein char-

acteristic of retro-elements. This was located in the59-most 973 bp that is in the opposite orientationfrom the b1 coding region (Fig. 2B). Although thisregion of gag homology has similarity at the aminoacid level to several high copy retrotransposons inmaize, the Grande element (accession no. X97604 andX97605), and elements in the adh1 flanking region(accession no. AF123535; SanMiguel et al., 1996), thelack of significant identity at the nucleotide levelprecludes the sequences in B-Bolivia from belongingto any identified maize retrotransposon family.

To better place the B-Bolivia insertion sequencewithin the context of these other gag sequences, weused the PROTPARS program of the PHYLIP pack-age to produce a phylogenetic tree of the B-Boliviainsertion sequence with the 19 gag proteins identified(“Materials and Methods”). The results of this anal-ysis, shown in Figure 3, demonstrated that theB-Bolivia insertion sequence is related to gag proteinsfrom other plant retrotransposons, but has signifi-cantly diverged from its nearest relation identified todate, the Sorghum bicolor Retrosor element (gi:4378066). These results strongly suggest that the se-quences in B-Bolivia that are absent from B-Peru rep-resent a retrotransposon- or retrotransposon-relatedsequence.

Several different classes of retrotransposons havebeen found in maize. Elements like Bs1, B5, G, Hop-

Figure 2. Structures of the b1 alleles. A, Restriction maps compar-ing B-Bolivia with the previously characterized B-Peru and B-Ialleles. The probes used in DNA-blot analyses are indicated oneach of the maps, and the gray boxes indicate the regions theyhybridize to. For clarity, only a subset of the mapped sites areshown. Restriction sites are BamHI (B), BglII (G), HindIII (H), andSpeI (E). B, The regions of the two alleles that have been cloned andcompletely sequenced are indicated by the brackets underneath thestructures. The sequences that are homologous between these twoalleles and other b1 alleles are shown in black, and the exons areindicated as large boxes and numbered. The upstream sequences ofB-Peru that are homologous to regions in B-Bolivia but not sharedwith other b1 alleles are shown in light gray, and the 534-bp repeatsare indicated by black arrowheads. The number above the B-Perusequence shows the end of the proximal 534-bp repeat sequenceand the number above the B-Bolivia sequence shows the point ofdivergence between the two alleles. The putative insertion inB-Bolivia is represented by open boxes that are separated by a gapindicating that the intervening distance and the orientation of thesetwo sequences has not been determined. The region labeled gagindicates sequences homologous to the gag proteins of variousretrotransposons. The arrow below gag indicates the direction ofthe reading frame.

Figure 3. Parsimony phylogenetic tree of gag protein homologs ofthe B-Bolivia upstream sequence. This phylogenetic tree representsthe consensus parsimony tree from analysis with 500 bootstrap rep-lications. The labels indicate the GenBank gi number for that proteinsequence and the two-letter abbreviation for the species from whichthe sequence derives. At, Arabidopsis; Os, Oryza sativa; Sb, Sorghumbicolor; Zm, Zea mays. Sequence 1363528 is from the Zeon-1element (Hu et al., 1995), all of the other putative proteins are fromuncharacterized retrotransposon-like elements. The branch lengthsindicate evolutionary distance and were determined from the FITCHprogram of the PHYLIP package (for details, see “Materials andMethods”). The bar labeled 20% change indicates the distance pro-duced by a 20% difference in sequence identity. Labels that appearto be at nodes of the tree are sequences that have very short distancesfrom the presumed ancestral sequence. Except for three of the nodesin the Arabidopsis clade, all branches are supported in greater than75% of the trees.




scotch, and Stonor are relatively low copy elementsassociated with genic sequences (Johns et al., 1985;Varagona et al., 1992; White et al., 1994). In contrast,there are several families of retrotransposons inmaize that have very high copy numbers and makeup a significant fraction of the maize genome (Ben-netzen et al., 1994; SanMiguel et al., 1996). To deter-mine the approximate copy number of the sequencesadjacent to B-Bolivia, we blotted known amounts ofthe cloned B-Bolivia upstream sequence and knownamounts of genomic maize DNA from several dis-tinct stocks on a slot blot. We probed the resultingslot blot with the cloned B-Bolivia sequence. Afterquantifying the blot, we found that 3 mg of maizeDNA had approximately 6 times the number ofcounts as did 10 ng of the cloned DNA from whichthe probe was made (Fig. 4). This intensity differenceprovides an estimate of approximately 38,000 copiesin the 2,500 megabase maize genome (“Materials andMethods”). Because the comparison of the genomicDNA signal was to a DNA sample that exactlymatches the probe sequence, the actual number ofcopies in the maize genome could be higher due tosequence divergence. The 38,000-copy number isvery similar to that estimated for the 9-kb Opie ret-rotransposon (SanMiguel et al., 1996) and severalother retrotransposons found in maize (Bennetzen etal., 1994; SanMiguel et al., 1996). However, the se-quence in B-Bolivia is clearly not any of these previ-ously described high copy retrotransposons.

B-Bolivia Has a Large Insertion or DNA RearrangementRelative to B-Peru

We next set out to define the size of the putativeinsertion in B-Bolivia. We used several probes to maprestriction enzyme sites in the upstream region ofB-Bolivia and B-Peru. By hybridizing with a probenear the 59 end of the transcribed region (550b;Patterson et al., 1995), we mapped several sites up-stream of the start of transcription in B-Bolivia (Fig.5). We next probed the same blot with a probe lo-cated 2.5 kb upstream of the start of transcription in

B-Peru (BIu4; Fig. 2A). This analysis did not revealthe size of the insertion as none of the restrictionenzymes tested yielded fragments that hybridized toboth the transcribed sequence and upstream se-quence probes.

Mapping of the regions farther upstream than theBIu4 probe revealed that B-Bolivia and B-Peru sharethe same pattern of restriction fragments with eightdifferent enzymes (“Materials and Methods”). Fur-thermore, PCR cloning of the region of B-Bolivia in-cluding the BIu4 probe and flanking B-Peru-like se-quence, confirmed that this region of B-Boliviamatched the sequence in B-Peru between 1 and 3 kbupstream of the start of transcription (left bracket inFig. 2B, underlined in Fig. 5). This region of B-Boliviais almost identical to the B-Peru sequence and con-tains one of the three 534-bp direct repeats found inthe B-Peru upstream region.

Having mapped the upstream sites, we could nowdetermine the location in B-Bolivia of sites down-stream of the BIu4 probe. We found that all of therestriction sites within 1.5 kb downstream of the BIu4probe in B-Bolivia were identical to the sites in B-Peru(data not shown). In contrast, all restriction enzymeswith sites farther than 1.5-kb downstream of BIu4produced different sized fragments, indicating thatthe sequences further than 1.5 kb downstream of theBIu4 probe were quite different in B-Bolivia relative toB-Peru (data not shown). Comparison of restrictionsites mapped upstream of the transcribed region (us-ing the 550b probe) and downstream of the BIu4probe, revealed that the two maps do not overlap (Fig.5). Assuming the full 2.5 kb of B-Peru-like sequence isstill found between these probes, there is approxi-mately 4.5 kb of non-B-Peru homologous sequencebetween the BIu4 probe and the HindIII site. Addingthis to the approximately 6 kb of non-homologoussequence between the 550b probe and the BglII sitegives a minimum size of 10.5 kb for the “insertion.” Itis also possible that a more complex DNA rearrange-ment is responsible for the juxtaposition of the retro-transposon sequences next to B-Bolivia.

Figure 4. Slot-blot analysis of the copy number of the putativeretrotransposon from B-Bolivia. The two columns on the left andmiddle are duplicates. The bracketed slots contain the indicatedquantity of unlabeled cloned B-Bolivia upstream DNA (a 1.3-kbregion) diluted into 3 mg of genomic carrier DNA (from petunia). Thebottom most slots of the first two columns contain 3 mg of maizegenomic DNA from the K55 B-Bolivia line (K606). The column onthe right contains the indicated amount of maize genomic DNA fromvarious stocks with the following b1 alleles: K606, B-Bolivia; J202, arecessive b1 allele; 1479, B-Gua31; 1490, B-marker ; and 1527, B-Peru.

Figure 5. Restriction map of the upstream region of B-Bolivia. Exons1 to 3 are numbered, and the B-Peru homologous sequences areindicated in the same manner as in Figure 2B. The labeled squareended lines indicate the BIu4 and 550b probes. The restriction sitesare abbreviated as follows: BamHI (B), BclI (C), BglII (G), EcoRI (R),HindIII (H), PstI (P), SpeI (E), and XbaI (X). The lines above and belowthe B-Bolivia map that end in white circle represent the regions ofB-Bolivia that have the same restriction map as B-Peru.




Transient Transformation Assays Reproduce theQuantitative Difference between B-Peru and B-BoliviaAleurone Pigment

One difference between B-Bolivia and B-Peru kernelpigmentation is that even the darkest B-Bolivia ker-nels are less pigmented than B-Peru kernels. A quan-titative transient transformation assay was used todetermine if the 22,100 B-Bolivia upstream regionproduces a different level of aleurone expression rel-ative to the 22,500 B-Peru upstream region that waspreviously studied (Selinger et al., 1998). In this assaythe B-Bolivia or B-Peru upstream sequences werefused to the reporter gene, firefly luciferase, and theconstructs were introduced into maize aleurone cells(Selinger et al., 1998). To normalize for transforma-tion efficiency, the test promoter:luciferase constructswere co-transformed with either a b-glucuronidase(GUS) reporter gene construct or a Renilla luciferaseconstruct both driven by the cauliflower mosaic virus(CaMV) 35S promoter. All luciferase values werenormalized to one of these transformation controls.We found that the 2.1-kb B-Bolivia upstream se-quence generated only 42% of the aleurone expres-sion produced by the 22,500 B-Peru construct, towhich we scaled all of our results (Fig. 6). This resultindicated that either some part of the B-Bolivia se-

quence reduced aleurone expression or some part ofthe B-Peru sequence that is missing from B-Boliviaenhanced aleurone expression. Our results discussedbelow suggest both types of events may be operating.

We had shown previously that 59-promoter dele-tions of B-Peru to 2176 had the same aleurone ex-pression in the transient assay as the 22,500 con-struct (Selinger et al., 1998). To determine if the singledifference in this region between the two alleles, afour base insertion in B-Bolivia relative to B-Peru justdownstream of the TATA box, had any effect onexpression, we compared a 2176 B-Peru promoterconstruct with the comparable 2180 B-Bolivia con-struct. Both constructs gave the same level of expres-sion, which was comparable with the level of expres-sion produced by the 22,500 B-Peru construct (Fig.6). This result indicated that the four base insertionimmediately downstream of the TATA box inB-Bolivia does not affect expression and that theB-Peru homologous sequences of B-Bolivia are fullycapable of driving levels of aleurone expressionequivalent to that of B-Peru.

A 33-bp Sequence from B-Bolivia ReducesAleurone Expression

The results with the 2180 B-Bolivia construct sug-gested that some other part of the 2,100 bp ofB-Bolivia upstream sequence was responsible for thereduced aleurone expression. To determine if part ofthe 1.5 kb of novel B-Bolivia sequence is responsiblefor the reduced expression we analyzed several de-letion derivatives (Fig. 6). We used the two XhoI sitesto produce three deletion derivatives. The smallestconstruct, a deletion to the XhoI site at 2564(2564BB:luc), produced a reduced expression levelthat was very similar to that of the 22,100 construct.The deletion of the 900 bp between the XhoI sites at21,454 and 2564 (dZ-BB:luc) also produced a level ofexpression that was similar to that of the 22,100BB:luc construct, corroborating the results from the2564 deletion and suggesting that the 33 bp betweenthe 2564 XhoI site and the start of the B-Peru homol-ogous sequence at position 2531 was sufficient toreduce aleurone expression (Fig. 6). However, dele-tion to the first XhoI site (21,454BB:luc) producedluciferase expression equivalent to the 22,500 B-Peruconstruct (Fig. 6). Because of these conflicting results,we decided to test the ability of the 33 bp between2564 and the B-Peru homologous sequences at 2531to reduce the expression of the 2176 B-Peru promot-er:luciferase construct. We tested a construct with asingle copy of the 33-bp sequence subcloned up-stream of the 2176 B-Peru:luc, and found that thissingle copy of the 33-bp sequence reduced aleuroneexpression to 40% (Fig. 6). Thus, the effect of the 33bp of sequence accounted for the reduction seen inthe 22,100 B-Bolivia construct. In addition, the reduc-tion to 40% in the transient assay correlated nicely

Figure 6. Deletion analysis of the B-Bolivia upstream promoter prox-imal region. The constructs shown represent promoter fusions toluciferase (shown as an open arrow labeled Luc) with the maizeAdh1 Intron1 sequence (labeled “a”) as a leader sequence. Details ofthe reporter gene are in “Materials and Methods.” Each construct wasintroduced into maize aleurones along with either a GUS or Renillaluciferase expressing control to normalize for transformation effi-ciency. The Luc/control activities were normalized to results ob-tained with the 22,500 B-Peru construct (100%; Selinger et al.,1998). The results were from at least five independent transforma-tions and are given with the SE of measurement. The XhoI sites (Z) arelabeled on the 22,100BB:luc construct and dashed lines indicate thesame sites in the deletion constructs. The sequences found inB-Bolivia that are not present in B-Peru are indicated by the un-shaded boxes. The bracket under the constructs indicates the 33 bpof B-Bolivia-specific sequence that is in common between five of theconstructs and that appears to contain a negative regulatory element.




with the approximately 30% level of anthocyaninpigment found in the darkest B-Bolivia seeds relativeto B-Peru seeds (data not shown). However, the re-sults with this chimeric B-Bolivia:B-Peru construct donot explain why the 33-bp sequence, which is presentin the 21,400 deletion construct, did not reduce ex-pression in that context.

The Region of the B-Peru Aleurone-Specific PromoterThat Is Missing from the Promoter Proximal Region ofB-Bolivia Contributes to Aleurone Expression

Besides the sequence that is unique to B-Boliviarelative to B-Peru, B-Bolivia is missing part of the534-bp repeated sequence that is found in B-Peru. TheB-Peru promoter region contains three identical534-bp direct repeats. All of the important aleuroneregulatory sequences are found in a single 534-bprepeat sequence (Selinger et al., 1998) and a B-Perudeletion derivative allele with a single 534-bp repeathas the same expression and stability as the nativeB-Peru allele (Harris et al., 1994). In the analysis of theB-Peru promoter, the 2710 and 2176 promoter re-gions produced equivalent expression in transienttransformation experiments (Selinger et al., 1998). InB-Bolivia, the aleurone-specific promoter consists of464 bp of the 534-bp repeat sequence; the distal 70 bpof the 534-bp repeat is either deleted or located else-where due to the insertion or rearrangement. Thecontrast between the stable expression of B-Peru andthe variable expression of B-Bolivia suggested thehypothesis that the 70 bp of the repeat that are miss-ing in B-Bolivia might have a quantitative effect onaleurone expression or the stability of thisexpression.

To determine if the 70-bp region might containregulatory elements important for aleurone expres-sion, we tested whether the presence of this regioncould suppress mutations in two critically importantregions in the first 176 bp of the promoter. We hadpreviously identified and characterized the E1 andE2 regions of the B-Peru aleurone-specific promoter(Selinger et al., 1998). In the context of the 2176B-Peru promoter, mutation of E1, which correspondsto the 2120 to 2109 region, results in a loss of ex-pression to 17%, whereas mutation of E2, which islocated between positions 296 and 285 results in areduction of expression to 7% (Fig. 7). When thesemutations were made in the context of the 2600B-Peru promoter and tested in the aleurone transienttransformation assay, expression was 40% and 60%,respectively, for E1 and E2 (Fig. 7). Importantly, wehad previously shown that a 2559 promoter con-struct carrying the E2 mutation had the same 7%level of expression seen in the2176 promoter (Fig. 7).These results suggest that the sequences in B-Perubetween 2600 and 2176 contribute to aleurone ex-pression and specifically localizes the element(s) crit-ical for the suppression of the E2 mutation to 31 bp

between 2600 and 2559, which is within the 70 bpmissing in B-Bolivia.

Transgenic Plants Containing the 2.1-kbUpstream Region of B-Bolivia Can ProduceAleurone-Specific Expression

Previous studies with the 22,500 bp region ofB-Peru had demonstrated that these sequences couldconfer aleurone but no plant expression. To deter-mine if the cloned upstream region of B-Bolivia couldreproduce the aleurone and plant expression pheno-types characteristic of this allele, we produced trans-genic maize plants carrying the B-Bolivia construct.We ligated the 2.8-kb B-Bolivia clone, which has 2.1kb of upstream sequence with the cloned B-I codingregion to produce a full length reconstruction of agenomic clone of the B-Bolivia upstream and B-I cod-ing region (Fig. 8A). We generated transgenic plantscontaining this construct by cobombarding thepBB2100 plasmid with a selectable marker, the bargene driven by the CaMV 35S promoter, into imma-ture embryos. We regenerated plants from 51 inde-pendent stably transformed lines that were resistant

Figure 7. The distal part of the 534-bp B-Peru repeat contains im-portant regulatory sequences. The structure of one of the B-Peru534-bp repeats (boxed black arrow) fused to the Adh1 intron 1 (a)and firefly luciferase (Luc) is shown at the top of the figure. The E1and E2 regions, which were previously shown to be important foraleurone expression (Selinger et al., 1998), are indicated by a grayand black box, respectively. The mutated versions of these twoelements are indicated by white boxes at the same position that E1 orE2 should be. These mutants were generated by substitution mu-tagenesis as previously described (Selinger at al., 1998). Each con-struct was introduced into maize aleurones along with either a GUSor Renilla luciferase expressing control to normalize for transforma-tion efficiency. The expression levels of the test constructs are indi-cated as percentages of the 22,500 B-Peru:luciferase expressionlevel set at 100%, after normalization to the internal control. Theresults for the 2559 B-Peru construct are taken from Selinger et al.(1998). Each construct was assayed in five independent transforma-tions and the mean value is given with the SE of measurement.




to the herbicide Basta, indicating that they were ex-pressing the bar gene. Plants from 14 of these inde-pendent lines expressed anthocyanin pigment inseeds when pollinated by a b1 tester stock, homozy-gous for functional alleles of all of the anthocyaninpathway genes except b1 and r1. The proportion ofBB2100 lines with seed pigmentation (27.5%) wassimilar to the proportion of B-Peru transgenic lineswith seed pigmentation previously isolated (36%,Selinger et al., 1998). DNA blots and/or PCR analy-ses on progeny derived from crossing the 14 differenthemizygous transgenic lines with non-transgenic b1tester demonstrated that, in at least nine of the 14lines, B-Bolivia transgene copies cosegregated as asingle locus (data not shown). One line clearlyshowed segregation of at least two transgenic loci(data not shown) and for the remaining four lines,insufficient data were obtained to clearly determinethat a single locus was segregating.

To determine if the BB2100 transgenic plantswould show the variable seed expression character-istic of kernels carrying the native B-Bolivia allele, wecounted the proportion of colored kernels on earsfrom T1 plants. For those lines with a single trans-gene locus, when each hemizygous transgenic plantis outcrossed, 50% of the progeny will receive thetransgene. If seed expression of the BB2100 transgene

was completely penetrant, like that of B-Peru, thenwe would expect all of the seeds from these nine linescarrying single locus BB2100 transgenes to be pig-mented, producing 50% colored kernels. Table IIIcontains a comparison of the proportions of coloredkernels in ears from all 14 lines with the nine linesverified to have a single transgene locus indicated.Two of the nine lines produced ears in which 50% ofthe kernels expressed pigment (lines VLC 40-16 and40-64, Table III). An example of one such ear isshown in Figure 8B. When we tested molecularly forthe presence of the BB2100 transgene in 20 colorlessseeds from one of these lines, we found that none ofthe kernels carried the BB2100 transgene. These re-sults suggest that, in these two lines, the BB2100transgene was completely penetrant, like B-Peru andthe B-Peru transgenic lines (Selinger et al., 1998). Inthe other seven verified single transgene locusBB2100 lines, less than 50% of the seeds were pig-mented. The ears were similar to that of the nativeB-Bolivia allele, in that these had reduced numbers ofcolored kernels and the colored kernels were gener-ally less pigmented (Fig. 8C). When we molecularlytested for the presence of the BB2100 transgene incolorless kernels from one of these lines with fewercolored kernels (VLC40-59), we found that 25% ofthe colorless kernels carried the transgene (5 of 20).

Colorless kernels that received the native B-Boliviaallele produced plants that had colored seeds in thenext generation. To test if the transgene loci behavedsimilarly, we grew and test-crossed transgenicBB2100 plants grown from colorless seeds from twoindependent lines known to have a single transgenelocus. We observed that transgenic plants grownfrom colorless kernels produced colored kernels inthe next generation in similar proportions to theirsiblings grown from colored seeds. These results in-dicated that in these lines the BB2100 transgenes inthe colorless kernels were not heritably silenced. Thisexpression pattern is reminiscent of the incompletepenetrance in seed expression seen in the nativeB-Bolivia stocks.

In addition to the incomplete penetrance of seedexpression, the native B-Bolivia allele normally failsto produce any pigmented seeds when transmittedthrough the pollen. We tested 11 of the 14 BB2100transgenic lines for seed pigment expression whentransmitted through pollen. For all 11 lines, coloredkernels were observed, regardless of which directionthe cross was performed with recessive b1 alleles.Data from two lines with multiple ears from crossesin both directions are presented in Table IV. Theseobservations indicate that none of the transgeniclines display parent of origin-specific expression.

BB2100 Transgenic Lines Produce NovelPlant Pigmentation

Six of the 14 BB2100 lines, in which the transgenewas expressed in the seed, displayed plant pigmen-

Figure 8. Phenotypes of B-Bolivia transgenic plants. A, Diagram ofthe BB2100 construct. The green block indicates B-Bolivia-specificupstream sequence, the red box with a black arrow represents thesequence that is homologous to the B-Peru aleurone-specific pro-moter, and the blue region indicates the exons (boxes) and introns(lines) of the transcribed region. This construct contains 2.1 kb ofupstream sequence from B-Bolivia together with exon1, intron1, andexon2 of B-Bolivia fused at the BamHI site in intron2 with theremaining genomic coding region of B-I. B and C, Two ears fromhemizygous BB2100 transgenic lines crossed by a b1 tester line. TheVLC 40-64 ear (B) has approximately 50% colored kernels, whereasthe VLC 40-59 ear (C) has approximately 30% colored kernels.D, The plant phenotype of a BB2100 transgene line (VLC 40-20).Note the phenotypic differences in the auricle (a), culm (c), and leafmid-vein (m) relative to Figure 1A.




tation (Table III; Fig. 8D). The phenotypes of these sixlines were quite similar to each other, but were dif-ferent from that of the native B-Bolivia allele. Theplant phenotype of the transgenic lines is essentiallyopposite of the phenotype of the native allele in threerespects (compare Fig. 1A with Fig. 8D). First, thetransgenic lines had strong expression in the auricletissue that separates the sheath from the leaf bladeand in the leaf mid-vein. Plants carrying the nativeB-Bolivia allele strongly pigment the sheath, butnever produce pigment in the auricle and rarely pro-duce weak pigmentation of the leaf mid-vein. Sec-ond, the transgenic lines produced relatively weakpigmentation of the culm, a tissue that is stronglyand uniformly pigmented by the native B-Bolivia al-lele. Third, three of the six transgenic lines producedpigment that was strong in the margins of the sheath.In contrast, the native allele produced no pigment insheath margins, even in plants with intense pigmen-tation in the rest of the sheath.

We considered several hypotheses to explain whyonly six of the 14 lines with seed expression showedplant expression and why this plant expression didnot mimic the native B-Bolivia allele. One possibilitywas that enhancers at the integration sites were in-fluencing the expression pattern. This seemed un-likely given that the chromosome position was dif-ferent in each line while the pattern of pigmentexpression was similar. In addition, in other experi-ments we generated many other transgenic lines con-taining either the B-Peru (Selinger et al., 1998) or B9 (KKubo, V. Chandler, personal communication) genomicclones, in which none of the plants exhibited any plantpigmentation. Thus, it is unlikely there are a fortu-itously high number of plant-specific enhancers in the

genome. A second possibility was that expression ofan endogenous b1 allele in the transgenic plants wasinfluencing transgene expression through an RNAsilencing mechanism (Jorgensen, 1995). This seemedunlikely as the presence or absence of plant pigmen-tation was stable and did not depend on the identityof the endogenous b1 allele or whether or not theendogenous allele was expressed in the plant (datanot shown).

Another possibility was that the lines lacking plantexpression have transgenes with promoter deletionsand are thus missing key regulatory sequences forplant expression. We used DNA-blot analysis and re-striction enzymes EcoRI and PacI, which flank thepromoter and coding region, respectively, to assessthe copy number of intact and partially deleted trans-gene copies in the 14 lines with seed expression. Theseresults, which are shown in Figure 9 and summarizedin Table III, indicate that nine of the BB2100 lines haveat least one intact copy, and one or more rearranged orpartially deleted copies of the transgene. Contrary tothe expectations of our hypothesis, four of the linesthat have plant color have no intact copies (Fig. 9). Sixof the nine lines with intact copies have no plantexpression, indicating that the presence of an intact2.1-kb upstream region fused to an intact coding re-gion is not sufficient for plant expression.

Four of the independent lines that produced strongseed pigmentation had no bands consistent with hav-ing an intact transgene copy, suggesting that theyhave at least one copy with an intact coding regionfused to promoter sequences sufficient for seed ex-pression. Because the PacI site is located a few basesbeyond the stop codon, most deletions of this site arelikely to result in the production of nonfunctional

Table III. B-Bolivia transgenic linesPlants hemizygous for each transgenic locus were outcrossed by b1 tester and the number of colored kernels determined.

TransgenicLine

% ColoredSeedsa x2 testb

PlantExpressionc

No. ofTransgeneCopiesd

SingleTransgene

Locuse

VLC 38-42 33 6 4 N/A 1 1 intact 1 6 N.D.VLC 39-2 47 6 5 N/A 2 1 intact 1 5 N.D.VLC 40-9 48 6 3 N/A 2 1 intact 1 3 N.D.VLC 40-12 41 6 5 9.91 3 10213 1/2 1 intact 1 5 YesVLC 40-16 48 6 3 0.0356 2 1 intact 1 3 YesVLC 40-20 44 6 3 N/A 111 0 intact 1 7 NoVLC 40-32 40 6 4 5.70 3 10216 2 1 intact 1 1 YesVLC 40-37 47 6 1 N/A 11 0 intact 1 2 N.D.VLC 40-38 44 N/A 2 ? intact 1 10 YesVLC 40-45 40 6 7 1.59 3 10226 2 1 intact 1 4 YesVLC 40-59 38 6 7 1.90 3 10225 111 0 intact 1 5 YesVLC 40-64 49 6 2 0.4752 11 1 intact 1 13 YesVLC 40-68 18 N/A 111 0 intact 1 3 Yes

a The mean and SD of kernel counts from at least four ears (except for the VLC 40-38 and 40-68 lines for which only one ear was availablefor analysis). b The probability from a x2 analysis that the actual seed counts are consistent with the expectation that 50% of the seeds arecolored. N/A, Not applicable. c Strong pigmentation of the auricle and leaf mid-vein tissues is represented by 111, moderate pigmentationby 11, light pigmentation by 1, very weak pigmentation by 1/2, and no anthocyanin pigment by 2. d As determined by DNA blots probedwith the 550-b probe. e Determined from segregating populations by either PCR- or DNA-blot analysis. N.D., Not determined. The VLC40-20line is clearly segregating at least two transgene loci, although only one may be functional.




proteins. The EcoRI site is located a few base pairsupstream of the BamHI site that is the 59-most end ofthe B-Bolivia upstream sequence (Fig. 9). Deletion ofthe EcoRI site and up to 1.5 kb of downstream se-quence may have little effect on seed expression be-cause the B-Peru homologous aleurone promoter se-quence would remain intact.

Although the complexity of the transgene arrays inmost lines made detailed characterization of theirpromoter structure difficult, PCR analysis was usedto determine the extent of the promoter deletions inthe two BB2100 lines with the fewest number oftransgene copies. These two lines, VLC 40-37 andVLC 40-68, had strong plant expression, but no intactcopies of the promoter region and only two to threepartially deleted copies (Table III; Fig. 9). We used aseries of upstream primers starting with the 531primer located at the upstream end of the B-Peruhomologous aleurone promoter region with addi-tional primers located further upstream and spacedapproximately every 300 bp (Fig. 9). Amplificationswere done using one of the upstream primers witheither of two downstream primers, the Sac primerlocated at the 59 end of intron 1 or the 532u primerlocated in the same region as the 531 primer butoriented in the opposite direction. Using the 531 andSac primers, we amplified a 550-bp fragment fromboth transgenic lines and from the native B-Boliviaallele indicating that both lines contained all of the531 bp of upstream sequence that is homologous tothe aleurone-specific promoter region of B-Peru (datanot shown). Additional reactions using upstreamprimers at 2700, 21,092, and 21,399 with one of thetwo downstream primers resulted in the amplifica-tion of the appropriately sized fragments from boththe VLC 40-37 and VLC 40-68 transgenic lines (datanot shown). PCR with an upstream primer at 21,682paired with the 2532 primer generated the expectedproduct with DNA from the VLC 40-68 line, but notthe VLC 40-37 line. Thus, the deletion in VLC 40-37begins somewhere between 21,399 and 21,682,

whereas the deletion in VLC 40-68 begins between21,682 and the EcoRI site at 22,100.

In summary, there was no clear correlation be-tween the presence of specific promoter sequencesand plant expression in the transgenic lines. A finalpossibility we considered is that another sequencewithin the transgene array may be contributing toplant pigmentation. This hypothesis is further devel-oped in the discussion.

DISCUSSION

Our results demonstrate that a DNA rearrange-ment that places sequences from a high copy numberretrotransposon adjacent to aleurone-specific pro-moter elements is associated with altered patterns ofexpression in B-Bolivia. Our finding that B-Boliviashares the same aleurone-specific promoter se-quences with B-Peru explains why both alleles areexpressed in the aleurone. However, it does not ex-plain the variability in the seed expression phenotypeof B-Bolivia relative to B-Peru. We hypothesized thatthe presence of a highly repetitive element adjacentto regulatory sequences required for aleurone expres-sion could be sufficient to produce the differences inseed expression between B-Peru and B-Bolivia. Toexamine this possibility, we produced transgenicplants containing 2.1 kb of the B-Bolivia upstreamsequence proximal to the start of transcription andperformed transient expression assays. Our resultsdiscussed below indicate that the sequences adjacentto the promoter can contribute to the variable pen-etrance of seed expression and to the reducedamounts of pigment, but they do not contribute tofemale-specific expression characteristic of the nativeallele.

In seven independent transgenic lines verified tohave a single transgene locus, the proportion of col-ored kernels was significantly less than 50%, indicat-ing that in these lines, the penetrance of expressionwas incomplete. In at least two of these lines, color-

Table IV. Transgenic lines do not show parent of origin differences in expressionPlants hemizygous for each transgenic locus were outcrossed with b1 tester and the no. of colored

kernels determined.

Stock Crossa Meanb N c Highd Lowd P e

% %

VLC 38-42 b13 32 4 39 29 –VLC 38-42 3b1 25 8 44 10 0.3337VLC 40-20 b13 44 6 48 41 –VLC 40-20 3b1 46 4 50 44 0.1859

a The direction of the cross that generated the ears that were scored is indicated as follows: 3b1, theBB2100 stock was crossed by b1 stock pollen; b13, the reciprocal cross in which BB2100 pollen wasplaced on b1 ears. b The average of the proportion of colored kernels on each ear is given. c Theno. of ears that were counted and used to determine the mean. d The proportion of colored kernelson the ears with the highest and lowest percentages in the sample. e The probability obtained by theANOVA method that the difference between the outcross by b1 pollen (3b1) and the outcross to b1 ears(b13) is due to chance variation.




less seeds that carry the transgene produce plantsthat express the transgene in progeny seed. This re-sult is similar to the behavior of the native B-Boliviaallele, in that colorless seeds that carry the B-Boliviaallele produce plants with indistinguishable expres-sion patterns and intensities in the next generationwhen compared with plants from colored seeds.However, the proportion of pigmented seeds in twoof the nine independent transgenic lines verified tohave a single transgene locus was approximately50%, indicating that in these lines 100% of the seedsthat received a transgene copy from the hemizygousparent expressed pigment. This proportion was sig-

nificantly higher than the proportion of colored ker-nels produced by either of the stocks that carry thenative allele. Furthermore, the copy number of thetransgene showed no correlation with the penetranceof seed expression. One possibility is that there is aspecific region within the B-Bolivia upstream se-quences that confers variability and the presence orabsence of this sequence is different in lines showing100% penetrance versus lines showing reduced pen-etrance. It is unfortunate that the number of copiesand the complex nature of the transgene loci makethis difficult to rigorously test.

Figure 9. DNA gel-blot analysis of the 14 BB2100 transgenic lines that produced seed pigmentation. Genomic DNA fromT1 transgenic plants was prepared and digested with PacI and EcoRI. The map below the blot indicates where these sites arewithin the construct that was introduced. After electrophoresis and blotting, the resulting blot was hybridized to labeled550b probe, washed, and hybridization detected using a Molecular Dynamics Storm 2000 system. The endogenoushomozygous B-615 allele produces a characteristic band labeled “B-615” in all lanes. Two bands representing the two r1alleles in the T1 heterozygotes are indicated by arrows labeled “r.” The expected size for an intact EcoRI/PacI digestedBB2100 transgene insertion is indicated by the arrow labeled “intact.” The lanes are labeled according to the “VLC” numberof the independent transgenic lines. The asterisks indicate the lines with vegetative plant pigmentation. The primers used todetermine the amount of B-Bolivia upstream DNA in two transgenic lines are indicated on the map.




Results from transient expression experiments,suggest that sequences within the putative retro-transposon insertion directly affect the amount ofaleurone pigment. Expression is reduced to 40%when a sequence normally located in the 33 bp of theinsertion proximal to the B-Peru homologous se-quences is placed immediately upstream of the 2176B-Peru promoter, a similar reduction to that seen inthe comparison of the 2.1 kb B-Bolivia upstream re-gion with the 176-bp B-Peru upstream region. How-ever, the promoter proximal region of B-Bolivia alsolacks sequences homologous to those between 2600and 2530 in B-Peru. The observation that the pres-ence of the 2600 to 2559 region reduces the negativeeffect of the two mutations (E1 and E2), suggests thatthe 2600 to 2559 element(s) and the 2120 to 284elements interact to regulate the expression of thealeurone-specific promoter. Thus, we hypothesizethat the retrotransposon insertion in B-Bolivia at 2530displaces important elements in the aleurone-specificpromoter. Transient expression studies and trans-genic analyses indicate these elements are not essen-tial, but we suspect their absence in the transgenes ordisplacement in the native allele contributes to thevariability of expression. An intriguing idea is thatthese sequences serve as a boundary element pre-venting other regulatory sequences from influencingthe promoter. The absence of the 2600 to 2559 se-quences in B-Bolivia may allow the retrotransposoninsertion in the native allele, and possibly other se-quences in the transgenic lines, to have a greaterinfluence on expression.

In addition to the incomplete penetrance and re-duced level of aleurone pigmentation in B-Bolivia,this allele also shows a parent of origin effect on seedexpression. All of the transgenic lines that expressedpigment in the seed, did so when transmittedthrough the male or female gametes. Thus the se-quences immediately upstream of the aleurone-specific sequences are not sufficient to impart theparent of origin effect on expression when the gene isin ectopic locations.

There are two possible ways to explain the lack ofpigment in kernels that inherit B-Bolivia from themale parent. The first explanation is that B-Boliviaexpression is sensitive to dosage, as the female con-tributes two doses and the male one dose to thetriploid endosperm, which gives rise to the aleurone.The second explanation is that B-Bolivia is epigeneti-cally imprinted such that it is completely silencedwhen transmitted through the male gametes,whereas, when transmitted through the female ga-mete, B-Bolivia is not silenced in all kernels. Al-though, due to technical problems, we have not beenable to directly test the dosage model, there are twoobservations that suggest that B-Bolivia is epigeneti-cally imprinted. The first is that there is no differencein the proportion of pigmented kernels or the amountof pigment between seeds with two and three copies

of B-Bolivia in the endosperm. The second is that wehave isolated a variant of B-Bolivia that does showseed pigment when transmitted through the maleparent (DA Selinger, VL Chandler, article in prepa-ration). The implication that B-Bolivia is an imprintedallele is particularly interesting in that there are onlya few well characterized imprinting systems in plants(for review, see Alleman and Doctor, 2000). In one ofthese systems, at the r1 locus in maize (Kermicle,1978), certain alleles, such as the R-r:std allele induceuniform pigmentation of kernels when passedthrough the female gametes, but induce a weaker,mottled expression when passed through the malegametes. At the Arabidopsis medea locus, expressionin certain ecotypes is solely from the maternallytransmitted allele in the endosperm and embryo tis-sues of the developing seed (Kinoshita et al., 1999;Vielle-Calzada et al., 1999).

In addition to the differences in aleurone expres-sion between B-Peru and B-Bolivia, the two alleleshave strikingly different patterns and levels of plantpigmentation. It is interesting that another allele ofb1, B-Gua31, isolated from an exotic land race (Negrode Chimaltenango) collected in Guatemala, has astrikingly similar pattern of plant pigmentation tothat of the native B-Bolivia allele. B-Gua31 lacks boththe aleurone-specific promoter sequences of B-Peruand B-Bolivia and the putative retrotransposon se-quences of B-Bolivia (Selinger and Chandler, 1999).Phylogenetic analyses of upstream sequences that areshared by all maize b1 alleles indicate that B-Peru,B-Bolivia, and B-Gua31 are closely related. However,the B-Gua31 and B-Bolivia alleles are the only mem-bers of the clade with strong vegetative plant pig-mentation phenotypes (Selinger and Chandler, 1999).One inference from this observation is that the puta-tive retrotransposon insertion does not carry theplant-specific regulatory elements that produce theplant expression seen in B-Bolivia, but rather, plantexpression is due to sequences outside of the retro-transposon region that differentiates B-Bolivia fromB-Peru. These sequences are presumably shared bythe phenotypically similar B-Bolivia and B-Gua31alleles.

Although the plant expression produced by the sixBB2100 transgenic lines is very similar between theindependent transgenic lines, it is quite differentfrom the plant expression of the native B-Bolivia al-lele. These differences in plant phenotype suggestthat if the sequences in the B-Bolivia upstream regionare producing plant expression, they are not behav-ing as they normally do in the native allele. Alterna-tively, another sequence in the transgene array maybe contributing to plant pigmentation.

Because these transgenic lines were produced byparticle gun bombardment, all of the lines have mul-tiple copies of the BB2100 construct along with theCaMV 35S promoter:bar gene construct, which servesas the selectable marker. An intriguing hypothesis is




that the enhancers that are part of the CaMV 35Spromoter of the selectable marker construct are in-teracting with the BB2100 transgene to induce plantexpression. The same selectable marker was used inthe generation of nine B-Peru transgenic lines, all ofwhich showed no plant expression. However, in theB-Peru transgenic lines, the potential insulator func-tion of the 2600 to 2559 sequences in the aleurone-specific promoter may have prevented this sort ofinteraction with the CaMV 35S enhancers. An alter-native explanation is that the highly repetitive natureof the B-Bolivia retrotransposon sequence in BB2100construct may produce very different interactionsbetween some of the transgene loci and the maizegenome, some of which may result in plantexpression.

As a model for the evolution of novel expressionpatterns, B-Bolivia reveals that the insertion in theupstream regulatory region of a high copy numberelement can change the expression pattern of a gene.Unlike the insertion in B-Peru that produces aleuronepigmentation, the insertion in B-Bolivia does not ap-pear to be carrying promoter elements that have beentranslocated from another gene. Instead it appearsthat insertion of this large, extremely high copy se-quence has altered the expression of the aleurone-specific sequences. Although these retrotransposonshave achieved extremely high copy numbers in anevolutionarily short time (SanMiguel et al., 1998),there is no evidence that any of them are still active,and they are very rarely found immediately next togenes (SanMiguel et al., 1996). In contributing to thereduced and unstable seed expression at B-Bolivia,this insertion may illustrate the consequences of hav-ing a large highly repetitive element near the pro-moter proximal and coding regions. Continued studyof the native B-Bolivia allele and various transgeniclines promises to define the roles played by differentsequence and chromatin structures in the control ofgene expression in plants.

Changes in the spatial and temporal expression ofgenes, especially genes encoding regulatory proteins,are likely to contribute to the evolution of new spe-cies and morphologies. A few genes have been iden-tified as major factors in conferring the morphologi-cal differences between maize and teosinte, its wildrelative (Beadle, 1939; Doebley and Stec, 1991, 1993).Recent work on one of these genes, teosinte branched 1,tb1, suggests that a change in expression is responsi-ble for the difference in the morphological pheno-types produced by the maize and teosinte alleles ofthis gene (Doebley et al., 1997; Wang et al., 1999).Similar types of changes in the cis-acting regulatoryregions of genes have been hypothesized to be re-sponsible for many instances of morphologicalchange during evolution (Doebley and Lukens, 1998).However, what the changes are and the molecularmechanisms that created the phenotypic variationfound at tb1 and many other genes are not well

characterized. It will be interesting to determine ifDNA sequence polymorphisms or DNA arrange-ments such as those observed at b1 are operating.

MATERIALS AND METHODS

Plant Materials

B-Bolivia in the K55 background was obtained from G.Neuffer (University of Missouri, Columbia) and B-Bolivia inthe GS background from George Sprague, Sr. (the Univer-sity of Illinois, Champaign/Urbana). Several b1 testerswere used, and all the testers carried recessive, nonfunc-tional alleles of b1 and r1, and functional, dominant allelesof the C1 regulatory gene, and the anthocyanin biosyn-thetic genes. The 414 stock resulted from a cross betweenthe K55 B-Bolivia line and a b1, r1, pl1-sr tester line. The1,470 stock is the result of a cross between the K55 and GSB-Bolivia that was then outcrossed to a b1, r1, Pl-Rhodestester.

Cloning B-Bolivia

Genomic DNA was extracted from an immature cob thatwas homozygous for the B-Bolivia allele, digested withBamHI, and fragments of approximately 2.8 kb, based onsize markers, were recovered by phenol extraction of themelted gel slices. DNA from the fraction showing thestrongest B-Bolivia-specific hybridization to the 550b probe(Patterson et al., 1995) was ligated into BamHI digestedlambda ZAP-Express arms and packaged using a Strat-agene XL-Gold packaging extract and plated (Stratagene,La Jolla, CA). Plaque lifts and hybridizations were per-formed using standard techniques (Sambrook et al., 1989)and the 550b probe. Positively hybridizing plaques werepicked, purified through a second round of plating, andhybridization, and the pBK phagemid containing the insertwas excised from the lambda ZAP vector according to themanufacturer’s instructions. The insert was subsequentlysubcloned into pTZ 18U for sequencing. This sequencealong with additional downstream sequence derived fromPCR experiments was deposited in GenBank (accession no.AF326577). Additional upstream B-Bolivia sequences wereobtained by PCR using oligos specific to B-Peru sequences(GenBank accession no. AF205801).

Sequence Analysis and Phylogenetic Analysis

The complete sequence of the 2.8-kb B-Bolivia clone wasused as a query for several searches using the BLASTN,BLASTX, and FASTX programs (Altschul et al., 1990; Pear-son et al., 1997). The potential protein sequence identifiedby these searches was then used for FASTA and TBLASTNsearches to identify homologous protein sequences andcoding sequences. The FASTA search returned 19 pre-dicted proteins with an E-value less than 0.01. A TBLASTNsearch with the same polypeptide sequence against thenucleotide database yielded 80 hits to nucleotide sequenceswith E-values less than 0.01. The 19 protein sequencesfound in the FASTA search with E() values less than 0.01




were aligned using the ClustalW program at the BCMserver (http://dot.imgen.bcm.tmc.edu:9331/multi-align/multi-align.html). The alignment was imported into theGenedoc program (Nicholas et al., 1997, available athttp://www.psc.edu/biomed/genedoc/). The N and Ctermini of the protein sequences were trimmed to removesequences that did not align well. The core region thataligned well for all sequences was then formatted for inputinto the PHYLIP package (Felsenstein, 1993) for phyloge-netic analysis. We used the PROTPARS program to pro-duce a tree based on the maximum parsimony method. Thetree was bootstrapped to generate confidence levels for allthe branches (Felsenstein, 1985). The same alignment thatwas used for the parsimony run was also put through thePROTDIST program using the “Categories” distance ma-trix to generate a table of evolutionary distances betweenthe sequences. The distance table and the consensus treewere then put into the FITCH program to generate a treewith branch lengths that reflect evolutionary distance.

Copy Number Determination

We prepared dilutions of unlabeled DNA, correspond-ing to 1.3 kb of the B-Bolivia upstream sequence from theupstream BamHI site to the point of divergence with B-Peruat 2532, into genomic Petunia DNA, which served as acarrier. Twenty-microliter samples containing 3 mg of Pe-tunia genomic DNA and 10, 1, 0.1, or 0.01 ng of the unla-beled B-Bolivia upstream sequence were diluted into 200mL of 0.5 m NaOH, 1.5 m NaCl denaturation solution andvacuum blotted onto a Hybond N1 membrane (Amersham-Pharmacia Biotech, Uppsala). In addition, 3-mg samples ofmaize (Zea mays) genomic DNA from a B-Bolivia line and 2-to 10-mg samples of genomic DNA from other maize lineswere similarly treated and blotted. The blot was then hy-bridized with a [32P]-labeled probe that was the same as theunlabeled cloned B-Bolivia upstream sequence used tospike the petunia DNA samples. After washing the blot,hybridization was quantified using a Storm 2600 Phosphor-imager. We used the following calculations to determinecopy number based on hybridization intensity. Ten nano-grams of 1.3-kb DNA equals 7.02 3 10E-9 copies of thesequence. Multiplying this number by 6, which is the dif-ference in intensity between the 10-ng probe signal and the3 mg of genomic DNA, and then dividing by the number ofmaize genomes in 3 mg of DNA (1.09 3 10E6) gives anestimate of approximately 38,000 copies in a single 2,500-megabase genome.

RFLP Mapping of B-Bolivia and B-Peru

Restriction mapping of the B-Peru and B-Bolivia alleleswas conducted using the same series of probes used to mapB-I and B-Peru (Patterson et al., 1995). The 550b probehybridizes to the region including intron1, exon2, and partof intron2 (Fig. 2A). The BIu4 probe hybridizes to theregion immediately upstream of the start of transcription inB-I, the E/G700 probe hybridizes to the region between21,600 and 2950 in B-I and the B’v1.6 probe hybridizes to

a region that is approximately 8 kb upstream in B-I (Fig.2A). All of these probes hybridize to regions tightly linkedto the B-Peru and B-Bolivia alleles. In the case of B-Peru, theisolation of recombinants has demonstrated that these se-quences are located much further upstream relative to theirposition in B-I (Patterson et al., 1995; M Stam, V Chandler,unpublished data). Band sizes were estimated by perform-ing linear regression on the log of the migration distance ofthe hybridizing bands compared with those of the knownsize standards run on the same gel. The E/G700 and B’v1.6probes, which are located much farther upstream of theBIu4 probe in B-Peru than in B-I (Fig. 2A; Patterson et al.,1995), had no polymorphic restriction fragments inB-Bolivia relative to B-Peru when tested with eight restric-tion enzymes that produce polymorphic fragments inB-Peru relative to B-I.

Construction of the BB2100 Transgenic Lines

The 2.8-kb clone of B-Bolivia containing 2.1 kb of up-stream sequence was subcloned between the BamHI sites ina 6-kb SalI clone of B-I that contained the complete tran-scribed region and 39 UTR, replacing the upstream regionof the B-I clone and the 59 most part of the transcribedsequence with B-Bolivia sequences (Fig. 8A). This constructwas introduced into maize plants using biolistic bombard-ment of immature embryos as previously described (Kozielet al., 1993; Selinger et al., 1998).

Characterization of BB2100 Transgenic Lines

The primary transgenic plants (T0) regenerated fromstably transformed callus were crossed to a b1 tester linethat carries functional alleles of all of the anthocyaninstructural genes and the regulatory genes c1 and pl1, whichare co-required with a functional b1 or r1 allele to generateanthocyanin pigment. The tester line carried nonfunctionalalleles of the b1 and r1 genes. The T0 plants, which were inthe Ciba-Geigy CG00526 inbred line, contained nonfunc-tional, recessive alleles of the regulatory genes c1, pl1, andr1, and a b1 allele that is weakly expressed in the plant, butthat does not color the seed. Because of the nonfunctionalc1 and pl1 alleles in the line used for transformation, wecould not assess pigmentation until the T1 generation.Transgenic lines that displayed seed color were furthercharacterized. DNA samples collected from leaves or im-mature cobs were used to characterize the copy numberand structure of the BB2100 transgene loci in those lineswith seed color. We used EcoRI, which cuts in thepolylinker of the pTZ 18U plasmid backbone and PacI,which cuts 10 bp beyond the stop codon. DNA blotsprobed with the 550b probe indicated the number of trans-gene copies, and by comparison with the size of the EcoRI/PacI digested BB2100 plasmid, the presence or absence ofintact transgene copies. To determine the number of trans-gene loci, colored and colorless seeds from each line wereplanted, DNA prepared and the presence of transgenecopies determined either by PCR using primers specific forthe transgene promoter and first exon region or by DNA




blots probed with the 550b probe. Lines were consideredsingle locus if there was no segregation of the transgenebands in purple seeds, and/or no colorless seeds carriedtransgene bands (minimum 7 samples, which equals a 99%confidence that a 50% probability event has not occurred).In some lines that had less than 50% colored kernels, wedid see transgenic bands in the colorless seeds, and thesewere scored as single locus if the individual bands did notsegregate when compared with purple seeds and if thefrequency of transgene positive, colorless kernels was closeto that expected from the frequency of colored kernels.

Transient Transformation Assays forAleurone Expression

Transient transformation of aleurone cells and luciferaseexpression assays were performed as previously described(Selinger et al., 1998). Briefly, expression constructs withB-Bolivia promoter fragments driving firefly luciferase ex-pression were introduced by biolistic methods into aleu-rone tissue along with a CaMV 35S promoter driven trans-formation control. The control plasmid was expressingeither GUS (Sainz et al., 1997) or Renilla luciferase (Lorenzet al., 1991; Selinger et al., 1998). Expression was deter-mined by normalizing the luciferase values to the transfor-mation control value and then dividing this number by thenormalized luciferase value of the 2.5-kb B-Peru promoterconstruct to generate a percent expression value. The B-Bolivia:luciferase expression constructs were produced bysubcloning the BamHI to SnaBI fragment (22,100 to 12,relative to the start of transcription) from the 2.8-kb BamHIB-Bolivia clone into BamHI/SnaBI digested pABPluc plas-mid, which was previously described (Selinger et al., 1998).This subcloning step replaced the B-Peru promoter se-quences in pABPluc with the B-Bolivia upstream sequence,which was located just upstream of the adh1 intron 1 se-quence and the firefly luciferase cDNA. Internal XhoI siteswere used to generate deletion derivatives pBB1400luc,pBBdZluc, and pBB564luc. To create the fusions with the2176 B-Peru promoter, oligonucleotides containing the se-quence of B-Bolivia between 2564 and 2530 were synthe-sized with engineered BamHI compatible 59 overhangs(Marshall University DNA core facility, Huntington, WV).The oligonucleotides were phosphorylated, annealed, andligated into the BamHI site of the BP176luc plasmid (Sel-inger et al., 1998). BP600 constructs were constructed byusing a 2600 B-Peru-specific oligo with an engineered 59-BamHI site together with the BP120A or BP96A oligos(Selinger et al., 1998) to PCR from 2600 to the mutant site.The engineered XhoI site of the linker-scan mutants to-gether with the BamHI site in the polylinker were used tosubclone the 2600 to 2120 or 296 fragment into the cor-responding 2176 B-Peru LS120 or LS96 luciferase construct.Other than the large deletions that were confirmed byrestriction analysis, all the constructs were confirmed bysequencing.

ACKNOWLEGEMENTS

We are grateful to G. Neuffer and G.S. Sprague, Sr. forproviding B-Bolivia stocks. We thank Susan Belcher for helpproducing and maintaining the transgenic lines, CatherineClay for help with DNA preparations and Southern blots,and Lyudmila Sidorenko and Teresa Lavin for commentson the manuscript.

Received November 20, 2000; accepted December 21, 2000.

LITERATURE CITED

Alleman M, Doctor J (2000) Genomic imprinting in plants:observations and evolutionary implications. Plant MolBiol 43: 147–161

Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ(1990) Basic local alignment search tool. J Mol Biol 215:403–410

Beadle GW (1939) Teosinte and the origin of maize. J He-redity 30: 245–247

Bennetzen JL, Schrick K, Springer PS, Brown WE, San-Miguel P (1994) Active maize genes are unmodified andflanked by diverse classes of modified, highly repetitiveDNA. Genome 37: 565–576

Coe EH Jr (1979) Specification of the anthocyanin biosyn-thetic function by B and R in maize. Maydica XXIV:49–58

Doebley J, Lukens L (1998) Transcriptional regulators andthe evolution of plant form. Plant Cell 10: 1075–1082

Doebley J, Stec A (1991) Genetic analysis of the morpho-logical differences between maize and teosinte. Genetics129: 285–295

Doebley J, Stec A (1993) Inheritance of the morphologicaldifferences between maize and teosinte: comparison ofresults for two F2 populations. Genetics 134: 559–570

Doebley J, Stec A, Hubbard L (1997) The evolution ofapical dominance in maize. Nature 386: 485–488

Felsenstein J (1985) Confidence limits on phylogenies: anapproach using the bootstrap. Evolution 39: 783–791

Felsenstein J (1993) PHYLIP (Phylogeny Inference Pack-age) version 3.5c. Distributed by the author. Departmentof Genetics, University of Washington, Seattle

Goff SA, Klein TM, Roth BA, Fromm ME, Cone KC,Radicella JP, Chandler VL (1990) Transactivation of an-thocyanin biosynthetic genes following transfer of b reg-ulatory genes into maize tissues. EMBO J 9: 2517–2522

Harris LJ, Currie K, Chandler VL (1994) Large tandemduplication associated with a Mu2 insertion in Zea maysB-Peru gene. Plant Mol Biol 25: 817–828

Hu W, Das OP, Messing J (1995) Zeon-1, a member of anew maize retrotransposon family. Mol Gen Genet 248:471–480

Johns MA, Mottinger J, Freeling M (1985) A low copynumber copia-like transposon in maize. EMBO J 4:1093–1102

Jorgensen RA (1995) Cosuppression, flower color patterns,and metastable gene expression states. Science 268:686–691




Kermicle JL (1970) Dependence of the R-mottled aleuronephenotype in maize on mode of sexual transmission.Genetics 66: 69–85

Kermicle JL (1978) Imprinting of gene action in maizeendosperm. In DB Walden, ed, Maize Breeding and Ge-netics. John Wiley & Sons, New York, pp 357–371

Kinoshita T, Yadegari R, Harada JJ, Goldberg RB, FischerRL (1999) Imprinting of the MEDEA polycomb gene inthe Arabidopsis endosperm. Plant Cell 11: 1945–1952

Koziel MG, Beland GL, Bowman C, Carozzi NB, Cren-shaw R, Crossland L, Dawson J, Desai N, Hill M, Kad-well S et al. (1993) Field performance of elite transgenicmaize plants expressing an insecticidal protein derivedfrom Bacillus thuringiensis. Biotechnology 11: 194–200

Lorenz WW, McCann RO, Longiaru M, Cormier MJ (1991)Isolation and expression of a cDNA encoding Renillareniformis luciferase. Proc Natl Acad Sci USA 88:4438–4442

Ludwig SR, Bowen B, Beach L, Wessler SR (1990) Aregulatory gene as a novel visible marker for maizetransformation. Science 247: 449–450

Ludwig SR, Habera SL, Dellaporta SL, Wessler SR (1989)Lc, a member of the maize R gene family responsible fortissue-specific anthocyanin production, encodes a pro-tein similar to transcriptional activators and contains themyc-homology region. Proc Natl Acad Sci USA 86:7092–7096

Nicholas KB, Nicholas HB Jr, Deerfield DW II (1997)GeneDoc: analysis and visualization of genetic variation.EMBNEW.NEWS 4: 14

Patterson GI, Kubo KM, Shroyer T, Chandler VL (1995)Sequences required for paramutation of the maize b genemap to a region containing the promoter and upstreamsequences. Genetics 140: 1389–1406

Pearson WR, Wood T, Zhang Z, Miller W (1997) Compar-ison of DNA sequences with protein sequences. Genom-ics 46: 24–36

Radicella JP, Brown D, Tolar LA, Chandler VL (1992)Allelic diversity of the maize b regulatory gene: differentleader and promoter sequences of two b alleles deter-mine distinct tissue specificities of anthocyanin produc-tion. Genes Dev 6: 2152–2164

Robbins TP, Walker EL, Kermicle JL, Alleman M, Della-porta SL (1991) Meiotic instability of the R-r complexarising from displaced intragenic exchange and intra-chromosomal rearrangement. Genetics 129: 271–283

Sainz MB, Goff SA, Chandler VL (1997) Extensive mu-tagenesis of a transcriptional activation domain identi-

fies single hydrophobic and acidic amino acids impor-tant for activation in vivo. Mol Cell Biol 17: 115–122

Sambrook J, Fritsch EF, Maniatis T (1989) Molecular Clon-ing, Ed 2. Cold Spring Harbor Laboratory Press, ColdSpring Harbor, NY, pp 2.60–2.112

SanMiguel P, Gaut BS, Tikhonov A, Nakajima Y, Ben-netzen JL (1998) The paleontology of intergene retro-transposons of maize. Nat Genet 20: 43–45

SanMiguel P, Tikhonov A, Jin Y-K, Motchoulskaia N,Zakharov D, Melake-Berhan A, Springer PS, EdwardsKJ, Lee M, Avramova Z et al. (1996) Nested retrotrans-posons in the intergenic regions of the maize genome.Science 274: 765–768

Selinger DA, Chandler VL (1999) Major recent and inde-pendent changes in the levels and patterns of expressionhave occurred at the b gene, a regulatory locus in maize.Proc Natl Acad Sci USA 96: 15007–15012

Selinger DA, Lisch D, Chandler VL (1998) The maizeregulatory gene, B-Peru, contains a DNA rearrangementthat specifies tissue-specific expression through bothpositive and negative promoter elements. Genetics 149:1125–1138

Stadler LJ, Neuffer M (1953) Problems of R gene structure:II. Separation of R-r elements (P) and (S) by unequalcrossing over. Science 117: 471–472

Styles ED, Ceska O, Seah KT (1973) Developmental dif-ferences in action of r and b alleles in maize. Can J GenetCytol 15: 59–72

Varagona MJ, Purugganan M, Wessler SR (1992) Alterna-tive splicing induced by insertion of retrotransposonsinto the maize waxy gene. Plant Cell 4: 811–820

Vielle-Calzada J-P, Thomas J, Spillane C, Coluccio A,Hoeppner MA, Grossniklaus U (1999) Maintenance ofgenomic imprinting at the Arabidopsis medea locus re-quires zygotic DDM1 activity. Genes Dev 13: 2971–2982

Walker EL, Robbins TP, Bureau TE, Kermicle J, Della-porta SL (1995) Transposon-mediated chromosomal re-arrangements and gene duplications in the formation ofthe maize R-r complex. EMBO J 14: 2350–2363

Wang R-L, Stec A, Hey J, Lukens L, Doebley J (1999) Thelimits of selection during maize domestication. Nature398: 236–239

White SE, Habera LF, Wessler SR (1994) Retrotransposonsin the flanking regions of normal plant genes: a role forcopia-like elements in the evolution of gene structure andexpression. Proc Natl Acad Sci USA 91: 11792–11796




b-bolivia, an allele of the maize b1 gene with variable expression, contains … · b-bolivia, an...

Documents